Why is it so hard to count Twitter bots

In recent years, automated accounts have become more sophisticated and complex. Many fake accounts in part controlled by the people, as well as machines, or just add to the message written by real people (that Mentser calls “cyborgs accounts”). Other accounts are using tricks designed to avoid detection by humans and algorithms, such as quick tweets mark the “likes” and “dislike” or publish or delete tweets. And, of course, there are many automated or pavavtamatyzavanyh accounts, for example, to control many companies that do not actually hurt.

The algorithm uses Machine Learning Botometer to evaluate a wide range of public data linked to the account, not only the content of tweets, but when sending messages you need for the account, and so on to determine the probability that it is a bot, Although the algorithm is state of the art, Menczer says, “a lot of accounts now fall into a range where the algorithm is basically not very sure.”

Menzer and others say that finding boots is a game of cat and mouse. But they add that in the future it may become much more complicated, as spammers use algorithms that generate better persuasive text and agreed to conduct a conversation.

Twitter itself is better equipped to detect bots through machine learning because it has access to much more data about each account. This includes a complete history of the user’s activities, as well as the various IP addresses and devices they use. But Delip Rao, a machine learning expert who worked on spam detection on Twitter from 2011 to 2013, says the company may not be able to disclose how it works because it may disclose personal data or information that may be used to manipulate platform recommendations. system.

This week, Musk also quarreled with Parag Agraval, CEO of Twitter, over how easily the company could disclose its bot search methodology. On Monday, Agraval sent the thread explains how difficult the task. He said that private information stored on Twitter, can make calculations about the number of boots on the service. «FirstnameBunchOfNumbers without a profile photo, and strange tweets may seem like a boot or spam, but behind the scenes, we often see a lot of indications that this is a real man,” – he wrote in the subject line. Agrawal also he said that Twitter can not disclose the details of these assessments.

If Twitter is unable or unwilling to disclose its methodology, and Musk says it will not continue without details, the deal could remain pending. Of course, Musk is using this problem as a lever to negotiate a price reduction.

At this point, Musk seems unhappy with Twitter’s efforts to explain why finding bots isn’t as easy as he thinks. He responded to Agraval’s long topic on Monday simple message it seemed much more suitable for a bot than for a potential Twitter buyer: the only, smiling emoticon with cocoa.